AI amplifies whatever it runs on. Teach it your official strategy and it executes the official strategy — including in situations where your people would actually do something different. We teach organizations to understand how they actually operate, and build that into their AI.
Start a conversation →We work with you over 3–12 weeks to build a Constitutional Profile — a clear, honest model of how your organization makes decisions. It's part structured research, part education, part design. The profile becomes the foundation your AI systems run on.
Three engagements, depending on how deep you want to go:
Diagnostic
We map the gap between your stated operating principles and your actual decision logic. You leave with a clear picture of where the misalignment is — and why your AI deployments feel slightly off.
$15,000
3 weeks
Redesign
Everything in the Diagnostic, plus a living constitutional document your team can actually maintain — and updated AI configurations that run on your real operating logic.
$40,000
6 weeks
Full Profile
Everything in Redesign, plus ongoing calibration and a persistent system that holds the constitutional layer as your organization changes. Built to outlast any individual person's memory of how things work.
$75,000
12 weeks
The engagement teaches your team to read five layers that your strategy documents don't capture — and that your AI needs to understand to reason the way you do:
What you've actually done under pressure
The decisions made when your stated principles were costly to hold. This is where real operating logic lives — in the moments your official values got tested.
What your team never questions
The assumptions so deep they don't register as assumptions. The things that get treated as given because questioning them feels like questioning the premise of the business.
What you haven't tried because you assumed it wouldn't work
Constraints that feel real but have never been tested. The options that get closed off before anyone tries them — and that your AI is also closing off by inheriting the assumption.
Who actually makes the calls
Real versus formal authority. Where information bottlenecks live. The org structure your AI should reason from — not the one in the handbook.
What has and hasn't changed when it was challenged
Whether your operating principles hold under pressure or only when it's comfortable. The ones that haven't updated when the evidence asked for it — and why your AI is carrying those forward.
This kind of work is educational in a specific sense: it asks people to update how they think, not just what they think. The strategic plan says one thing; the organization does another; the AI picks up the gap. Closing the gap means being willing to look at the gap honestly.
Before every engagement, we run a two-hour session to understand whether the leadership team is ready to do that. It's included, and it determines what kind of work is possible. If the answer is that the organization would benefit more from a different kind of engagement, we'll say so directly.
Before building this site, we ran three independent analyses of our own operating logic — same source material, different sessions — and had a judge synthesize between them. We found the same gap we find in clients: the theory had outrun the deployment. The site is the correction. We practice what we consult.
A founder had built a complete financial exchange solo. Killed by Brazilian regulation. Pivoting to prediction markets. Pre-call reading: "he treats distribution as incompatible with his identity, pivots when a hard wall appears, and repeats." Prediction markets had the same regulatory profile. The pivot wasn't an escape. The call confirmed it. He'd known — and hadn't named it.
We take a small number of engagements at a time. Each one makes the instrument more accurate. If something on this page describes a problem you're working through, reach out.
Take ten minutes. If these point at something familiar, the conversation will be useful.
When your AI system encounters a situation your instructions don't cover — what does it do?
When did you last update the document your AI runs on to reflect what actually changed in how you operate?
After a significant AI-assisted session, what's the decision record — what was concluded, why, and what would change it?
How far is the gap between how your organization describes itself and what it actually prioritizes when things get hard?
Can your leadership team make a genuine case against their own current strategy — one that would require changing the framework, not just updating a belief within it?
Question five is the one that matters most. If it produced discomfort, that's the signal. The engagement has already started.
A first call is 30 minutes. You describe where the problem is showing up; we describe whether this kind of work addresses it. If it does, we'll walk through what an engagement looks like.
Epiplexia — Constitutional Consulting
Andy Trattner · [email protected]